You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
36 lines
3.2 KiB
Plaintext
36 lines
3.2 KiB
Plaintext
<!--200810-->
|
|
<h1>slimeAI - state transition diagram </h1>
|
|
december 10, 2020<br>
|
|
#ai #programming<br>
|
|
<br>
|
|
<br>
|
|
blessfrey's slimes use finite state machines. <b>Finite state machines</b> (FSM) are models with a finite number of states that either decide to generate output or transition to a new state based on input. They can only be in one state at a time. <br>
|
|
<br>
|
|
<br>
|
|
<h2>why use a finite state machine? </h2><br>
|
|
Because they are simple and flexible, while still allowing for interesting behavior. FSMs reflect our own mental model of how slimes behave (attack mode, wander mode, sleep mode, etc). These different modes or states are written as manageable chunks that easy to code, debug, update, and refactor. It's also efficient with processor time, since the behavior is abstracted to basic conditionals. <br>
|
|
<br>
|
|
A state can be as broad or specific as you need, so long as it's granular. So they can be more easily debugged and swapped out with one another, it's a good idea to have one entry point and one exit point for all your states.<br>
|
|
<br>
|
|
Of course, since FSMs can be used for anything with states and transitions. You can think of specific menus, mechanics or even your entire game in terms of states. I'm just using them here for understanding the AI for my slime monster. <br>
|
|
<br>
|
|
<br>
|
|
<h2>designing AI </h2><br>
|
|
Planning ahead is extra important, since any dead-ends will halt your slime. I used a state transition diagram to design the slime's AI. In the diagram, circles represent states and arrows represent transitions, triggered by some kind of event. I marked the initial state with a star, but Buckland recommends a double circle. <br>
|
|
<br>
|
|
The slime is a simple creature. It idles, chases anything that gets too close, and attacks until its target dies. Here's the diagram version of that:<br>
|
|
<br><center>
|
|
<img src="/static/img/ent/slimeai_diagram.png" alt="(image: a state transition diagram for slime ai.)" width="500" height="250"><br>
|
|
</center>
|
|
<br>
|
|
The states are Idle, Choose Target, Attack, and Move to Target. The initial state is Idle. <br>
|
|
<br>
|
|
If there's a valid target, the slime enters Choose Target. If there's no longer a valid target, it returns to Idle. If it successfully gets a target, it begins moving to target. If the target becomes invalid, it returns to Idle. Once the target is in attack range, it enters Attack. If the foe moves out of range, it returns to Move to Target. If the target becomes invalid, it returns to Choose Target. And the cycle continues. <br>
|
|
<br>
|
|
No matter how you follow the arrows, there are no dead-ends and all states can be reached eventually, so it'll work okay when it's implemented as code. <img src="/static/img/emo/heart.gif" alt="<3"><br>
|
|
<br>
|
|
<br>
|
|
<h2>reference </h2><br>
|
|
<a href="https://www.goodreads.com/book/show/161139.Programming_Game_AI_by_Example"><b>Programming Game AI by Example by Mat Buckland</b></a>. It's an easy to understand introduction to game math + AI concepts. The code examples are in C++, but they are still useful for understanding the underlying principles. I didn't even know where to begin before I started reading and it got me pretty decent groundwork for AI creatures with different behavior that reacts to the players.<img src="/static/img/emo/heart.gif" alt="<3"> <br>
|
|
<br>
|